Challenges in NVH analysis

LinkedIn +

The demand for more focused NVH evaluation is driving the need for innovative measurement techniques. We learn of some of the latest tools and equipment available

For the acoustic engineer, the task of selecting and learning how to harness available NVH test and analysis tools has become an immense burden. Alongside that, NVH analysis has been brought into play much earlier in the design cycle through the integration of finite element analysis with computer-aided design.

Actran acoustics software, developed by Free Field Technologies, is used to predict the acoustic, vibro-acoustic and aero-acoustic behavior of vehicle subsystems and components. In recent years, experts at the company have implemented several key improvements in the technology.

Jean-Louis Migeot, founder and CEO, explains, “The first big improvement relates to calculation time. A computer may have to run for hours or even days to solve a problem, but engineers want to have an answer within the day. We have worked a lot on improving the computational performance of our software and in Actran version 16 calculation time is not only much shorter but the software can also tackle larger problems up to higher frequencies.

“Part of that work has been in making sure our algorithms are as efficient as possible, which is mainly based on mathematical research, but also in ensuring that our software runs on all types of machines. Whatever the computational power available, the software needs to be able to harness it and to produce results as quickly as possible.”

Maintaining a balance between accuracy and speed, however, can be challenge, as Migeot notes: “It’s about providing a relevant answer to the problem in a reasonable amount of time. If it’s too accurate it’s impossible to solve, but if it’s too fast it’s inaccurate.”

Achieving high levels of accuracy not in absolute terms, but in relative terms is important. “If you compare two material designs, for example, and identify a 2dB difference between them, you also want to see a 2dB difference when you physically design and test the system. We are now reaching that kind of accuracy,” explains Migeot.

“But the software is one part and the user is another important part because he has the responsibility to create the right model. That’s why we spend a lot of time with our customers in making sure that we jointly understand the application, so that we can guide them in forming the question correctly. I always say that the software gives the right answer but it cannot assess whether the question is correct.”

Improving the productivity and ease with which models can be constructed in Actran has helped reduce overall calculation time. “This can be quite time consuming and in version 16 we have included internal meshing and automated meshing algorithms, which we’ve taken from MSC Software.” The vehicle’s trim package in particular is extremely difficult to model due to the increasing number of materials and their complexity. It’s been a central focus of R&D at Free Field Technologies and an area that the company has made major progress in. “We’ve done that by creating the right material models and finding the right way to blend the model of the individual components with the full model of the car,” Migeot says. “We’ve also added algorithms to describe the affects of flow and changing temperatures to enable the creation of models for as many different components and subsystems as possible.”

Meanwhile, suppliers are also helping to address the physical test needs of acoustic engineers. A team of engineers at Siemens PLM Software recently developed an improved method for frequency response function acquisition in vehicle NVH analysis. Traditional methods, involving a modal hammer or a shaker used to excite in three dimensions at interface connection points throughout the vehicle, are becoming problematic due to the compactness of modern vehicles restricting access to relevant points.

Steven Dom, automotive solutions manager at Siemens Industry Software, says, “Standard hammer or more modern shaker measurements are limited in application when three dimensional inputs to a location are required. One dimension is easy but the others much more difficult and often not achievable without some sort of added structure.”

Developed through a customer project, the Siemens solution consists of a small block with angled sides and a built-in 3D accelerometer in combination with Siemens shakers. “The method simply relies on the fact that the coordinates of a vector can always be transformed in an arbitrary coordinate system to the coordinates of that vector in another coordinate system by what is called a ‘coordinate system transformation’ – a simple mathematical operation that can easily be automated,” explains Dom.

It was validated on both simple plate structures and on complete car bodies through comparative studies with traditional methods, which showed equal results. “We also did the typical reciprocity checks to ensure the quality of traditional measurements.”

According to Dom, the new system is not only more practical but is also highly accurate and representative of the real world: “To a certain extent you could say it is more accurate than current methods because this block is very light and directs the applied forces through a single point on the surface. It’s not the same as a structure that you build on top of the surface to excite the tangential directions. Hitting such a structure introduces an additional, unwanted moment into the structure.

“The overall benefit,” he adds, “is clearly that with this method, measurements that would otherwise be very difficult to take are now possible. This does not mean that the current methods are bad, but the new one adds another tool to the box.”

In recent years acoustic holography – a technique used to estimate the soundfield near a source by measuring acoustic parameters away from the source via an array of pressure and/or particle velocity transducers – has been increasingly adopted.

Experts at Sorama, a spin-off company from the University of Technology Eindhoven in the Netherlands, have developed the world’s largest microphone array for both near-field and far-field measurements using acoustic holography.

“One of our design specifications in developing the system was that it should be scalable, so that an array can be built based on the requirement of the measurement and the available budget,” explains embedded systems and software engineer Wouter Ouwens.

“We chose to build an 8 x 8 microphone array that uses digital MEMS microphones spaced 2cm apart in both dimensions. These microphones are the same as those used in smartphones and are produced in large volume, so they cost only US$1 each. This is considerably cheaper than traditional analog microphones.”

To ensure the system is scalable, it was important to incorporate modularity into the design of both the hardware and the software. “In the hardware we have developed multiple, small basic building blocks, which are combined into a single array. The same principle is applied to our software, where it is divided into small reusable chunks and can run on different platforms,” explains Ouwens.

“A great benefit of this modularity is the testability of both the software and the hardware in the system. Each module in the system can be tested by itself, since the behavior of the sub-module is well defined.”

The basic array is interconnectable, so that multiple arrays can be combined into a single array. “To achieve that we have integrated local processing, local storage, power electronics and synchronization capabilities. The array can be connected to a computer via Ethernet or USB.”

By using a large array, it has been possible to drastically reduce measurement times compared with traditional systems. A single measurement can create data for the whole side of a vehicle in only a second. Transient behavior can also be observed and fully synchronized with thousands of microphones all at once.

The system is said to offer a much lower threshold and is user friendly for companies just starting to use sound cameras. The over-sensed arrays enable the user to use both far-field super-resolution beamforming methods of acoustic holography as well as Sorama’s near-field acoustic holography. Displacement accuracies in structures in the order of nanometers have already been proved in practice.

Developing the system was not without its challenges, Ouwens notes. “The array has 4,096 channels, which generate a data stream of 6GB per second. Processing this data in a reasonable time is difficult, so the hardware has a lot of local storage to save up to 10 seconds of data. Also, the hardware should be able to process all this data. It was a challenge to get the pre-processing up and running on the processor of the array itself.”

In the future, Sorama hopes to develop the system beyond the 4,096 channel array and build a ‘wall’ of microphones (Nx1,024 channels), so that even larger areas can be measured at one time. “However, the data streams for this would be even larger and currently we’re researching how we could handle that.”

Sidebar: no problem

Without the masking noise of a traditional combustion engine, NVH in transmissions of hybrid and electric vehicles can be a challenge to optimize. It’s an area that engineering consultancy Drive System Design (DSD) is well versed in.

“A lot of people tend to think that noise in transmissions is a very complicated problem to work out. It’s not really, as there’s a lot of existing math that makes it relatively straightforward. We’ve managed to solve dozens of these types of problems now, where people were stuck very close to production because NVH is one of the last things that gets solved,” says Alex Tylee-Birdsall, technical director at DSD.

In a recent example, experts at DSD managed to reduce the vibration of a transmission casing by adding ribs, which changed its damping characteristics.

“Ribbing is a well-known and successful technique for improving strength and stiffness. It is also an effective way of managing local vibration frequencies,” explains Tylee-Birdsall.

“Often we use optimization tools like OptiStruct (pictured below) to define the location and dimensions of ribs, although if you consider load paths as traveling from bearing to bolts, rib directions are relatively easy to define for strength. We have developed our inhouse tools to enable estimation of the number and dimensions of the ribs as a first pass approximation for NVH improvement prior to the use of topology optimization tools.”

The ribs (pictured bottom), which are now a regularly used design technique, are always tested the same way in both simulation and physical tests. “We place accelerometers at key identified locations. This provides us with the required data to evaluate the effectiveness of the ribs. In cases where we are looking to reduce the transfer of noise, rather than reduce the source, we must consider transfer functions from the accelerations to radiated noise in the vehicle.

“Simulation results so far have been very close to testing. I think we are surprised at how close we have got and we’ve been able to use simulation to reduce a number of noise problems.”

January 26, 2016

Share this story:

About Author


John joined UKi Media & Events in 2012 and has worked across a range of B2B titles within the company's automotive, marine and entertainment divisions. Currently editor of Automotive Testing Technology International, Crash Test Technology International and Electric & Hybrid Marine Technology International, John co-ordinates the day-the-day operations of each magazine, from commissioning and writing to editing and signing-off, as well managing web content. Aside from the magazines, John also serves as co-chairman of the annual Electric & Hybrid Marine Awards and can be found sniffing out stories throughout the halls of several of UKI's industry-leading expo events.

Comments are closed.